J Clin Oncol 27: by American Society of Clinical Oncology INTRODUCTION

Transcription

1 VOLUME 27 NUMBER 27 SEPTEMBER JOURNAL OF CLINICAL ONCOLOGY O R I G I N A L R E P O R T Effect of Exercise on Postmenopausal Sex Hormone Levels and Role of Body Fat: A Randomized Controlled Trial Evelyn M. Monninkhof, Miranda J. Velthuis, Petra H.M. Peeters, Jos W.R. Twisk, and Albertine J. Schuit From the University Medical Center Utrecht, Julius Center for Health Sciences and Primary Care; Comprehensive Cancer Center Middle Netherlands, Utrecht; Institute for Health Science, VU University, Amsterdam; and National Institute for Public Health and the Environment, Bilthoven, the Netherlands. Submitted August 27, 2008; accepted March 12, 2009; published online ahead of print at on August 17, Supported by Dutch Cancer Society Grant No. UU Presented at the 2nd International Congress on Physical Activity and Public Health, Amsterdam, the Netherlands, April 13-16, The support from the sponsor was unconditional, and the data collection, design, management, analysis, interpretation, and reporting were performed without their interference. The role of the sponsor was limited to approving the scientific proposal of the study; covering salary costs of study personnel, costs for the data collection, and costs for biochemical analyses. Authors disclosures of potential conflicts of interest and author contributions are found at the end of this article. Clinical Trials repository link available on JCO.org. Corresponding author: Evelyn M. Monninkhof, PhD, University Medical Center Utrecht, Julius Center for Health Sciences and Primary Care, STR 6.131, PO Box 85500, 3508 GA Utrecht, the Netherlands; e.monninkhof@ umcutrecht.nl. The Acknowledgment is included in the full-text version of this article, available online at It is not included in the PDF version (via Adobe Reader ) by American Society of Clinical Oncology X/09/ /$20.00 DOI: /JCO A B S T R A C T Purpose To examine the effects of a 1-year exercise intervention on sex hormone levels in postmenopausal women and whether any effects are mediated by changes in body fat composition. Methods We randomly assigned 189 sedentary postmenopausal women (age 50 to 69 years, body mass index of 22 to 40 kg/m 2 ) to an exercise intervention (n 96) or a control group (n 93). The intervention combined aerobic and strength training and comprised supervised group sessions and home-based exercises (a total of 2.5 h/wk). Between-group differences in sex hormone levels (at baseline and 4 and 12 months) were examined with generalized estimating equations. Results In total, 183 women (97%) completed the study. Overall, the exercise intervention did not result in favorable effects on sex hormone levels. Among women who lost more than 2% body fat, declines in all estrogens were not significantly different between exercisers and controls. Androgen levels decreased significantly in the exercise group who lost body fat compared with their peers in the control group. Furthermore, this study confirmed that fat loss was significantly associated with declines in postmenopausal estrogen levels. Although not significant, a similar trend was observed for the androgens. Conclusion This study confirms that fat loss is associated with changes in postmenopausal sex hormone levels and suggests that exercise may be effective in inducing favorable changes in these hormones. J Clin Oncol 27: by American Society of Clinical Oncology INTRODUCTION Most of the established risk factors for breast cancer, such as family history of the disease, early age at menarche, late age at menopause, late age at first childbirth, or nulliparity, are not, or not easily, amenable to intervention. Physical activity, however, is a modifiable lifestyle characteristic that has been associated with decreased breast cancer risk. 1,2 The causal mechanism is not clear, and it is unknown whether increasing physical activity at later age still reduces breast cancer risk. Sex hormones are mainly proposed as mediators for the observed association between exercise and breast cancer. 3,4 The evidence that estrogens contribute to breast cancer risk is strong and widely accepted. A reanalysis of nine prospective studies showed that a relatively high level of endogenous estrogens in postmenopausal women is associated with a two- to three-fold increased risk of breast cancer. 5 Postmenopausal women with higher androgen levels also have increased risk. 5,6 In cross-sectional studies, a low level of physical activity has been associated with 15% to 25% higher serum concentrations of estradiol, estrone, and androgens in postmenopausal women after adjustment for body mass index (BMI) 7-10 ; however, these results have not been observed in all studies. 11 This association might be partly explained both directly and also indirectly through accumulation of adipose tissue. 12,13 After menopause, adipose tissue is the primary source of estrogen production as a result of peripheral conversion from androgens by aromatase. 14,15 Compared with normal-weight postmenopausal women, obese postmenopausal women have higher blood concentration of estrogens and androgens and lower concentration of sex hormone binding globulin (SHBG). 10,16,17 Regular exercise represents an approach to regulate energy balance and to prevent accumulation of adipose reserves and consequently influences the production of estrogens. So far, one study has been published assessing the effects of exercise interventions on sex hormones in US women. 18, by American Society of Clinical Oncology

2 Exercise and Postmenopausal Sex Hormone Levels This randomized intervention study of 12-month moderate-intensity exercise in sedentary, overweight, postmenopausal women resulted in significant decreases in serum estrogens and androgens, but only in women who lost body fat. To assess this relationship in Dutch women who differ from US women in terms of weight and physical activity level, we designed the Sex Hormones and Physical Exercise (SHAPE) study (trial registration No. NCT ). We hypothesized that exercise would lower sex steroid levels mainly as a result of changes in total body fat. METHODS The SHAPE study is a randomized controlled trial examining the effects of a 1-year moderate- to vigorous-intensity exercise program on sex hormone levels among sedentary postmenopausal women. The study was approved by the ethical committee of the University Medical Center Utrecht, and written informed consent was obtained from each participant. The design of the SHAPE study has been described previously. 20 In short, we included 189 healthy sedentary postmenopausal women age 50 to 69 years. These women were recruited through a random selection of the female inhabitants of several municipalities in the Netherlands. Postmenopausal status was defined as having no menses for at least 1 year, 21 and sedentary was defined as engaging less than 2 hours per week in moderate sport and recreational activities and not adherent to the international physical activity recommendation. 19 Exclusion criteria included use of hormone replacement therapy, smoking, BMI less than 22 kg/m 2, corticosteroid use, cancer in the past 5 years, and diabetes mellitus or other endocrine-related diseases. Eligible women were enrolled by a study nurse. Random assignment was concealed and blocked on two categories of waist circumference ( and 92 cm; cutoff level was based on the median value reported in comparable women). 22 Intervention Women in the intervention group participated in a combined aerobic and strength training program over a period of 12 months. The program was organized in a way to induce fat loss and to reach a training stimulus in each person. The exercise prescription consisted of 1-hour supervised group sessions twice a week and an individual session of 30 minutes once a week. Further details have been described elsewhere. 20 Control participants were requested to retain their habitual exercise pattern. Exercisers and controls were asked to maintain their usual diet. Outcome Measurements The study participants visited the medical research unit at baseline and after 4 and 12 months. A study nurse performed measurements and collected information by questionnaires. Each visit, blood samples were taken. In addition, anthropometric measurements and dual-energy x-ray absorptiometry to measure body fat were performed. Also, medication use was assessed at each visit. Invitation letters mailed to women aged years (n = 6,200) Response to invitation letter (n = 1,799) Screened on eligibility criteria by phone (n = 1,360) Subject information mailed to eligible women (n = 310) Not eligible (n = 1,046) Refused to participate (n = 4) Invited for baseline visit (n = 208) Randomly assigned (n = 189) Refused to participate (n = 102) Not eligible (n = 19) Most important reasons: High blood glucose Low BMI Physically active lifestyle Fig 1. Flow chart of the inclusion, random assignment, and retention of the Sex Hormones and Physical Exercise (SHAPE) study participants. BMI, body mass index. Intervention group (n = 96) Compliant (n = 60) Attended < 70% of group sessions (n = 35) Dropped out of the study (n = 1) Control group (n = 93) Compliant to control status (n = 77) Not compliant (ie, started exercising, a weight loss program, or smoking) (n = 11) Dropped out of the study (n = 5) Blood available Baseline (n = 96) 4 months (n = 94) 12 months (n = 94) Analyzed (n = 96) Blood available Baseline (n = 93) 4 months (n = 87) 12 months (n = 88) Analyzed (n = 93) by American Society of Clinical Oncology 4493

3 Monninkhof et al Blood Samples Blood samples were drawn between 9:00 and 11:00 AM after an overnight fast and stored at 70 C. Serum concentrations of estradiol (total and free), estrone, estrone sulfate, testosterone (total and free), androstenedione, and SHBG were determined by laboratory technicians blinded to intervention status. All samples of one individual were analyzed in the same batch. Commercially available double-antibody radioimmunoassay kits were used (Diagnostic System Laboratories, Webster, TX, and ZenTech, Angleur, Belgium) for estradiol (DSL-4800), estrone (DSL-8700), estrone sulfate (DSL-5400), testosterone (DSL-4100), androstenedione (DSL-4200), and SHBG (IRMA CT). Kits for estrone, androstenedione, and testosterone showed a high relative validity in terms of ranking postmenopausal women. 23 The mean intra-assay coefficients of variation were as follows: 10.0% (estradiol), 14.0% (estrone), 6.4% (estrone sulfate), 9.8% (testosterone), 5.3% (androstenedione), and 12.8% (SHBG). Free estradiol and free testosterone were calculated using the measured values for estradiol or testosterone, SHBG, and an assumed constant for albumin. 24 For SHBG, estrone, and estrone sulfate, some women had undetectable levels at one of their visits ( 0.26 nmol/l, 1.2 pg/ml, and 0.01 ng/ml, respectively). At baseline, four women had undetectable estrone levels. At 4 months, one woman had undetectable SHBG levels, and two women had undetectable estrone levels. At 12 months, we found undetectable levels for one woman for SHBG, for three women for estrone, and for one woman for estrone sulfate. For the statistical analyses, the levels of these women were Table 1. Baseline Demographics and Clinical Characteristics of the Participants of the Sex Hormones and Physical Exercise Study Characteristic Exercise Group (n 96) Control Group (n 93) P for Difference Age, years.45 Mean SD Years since menopause.30 Mean SD Weight, kg.39 Mean SD Body mass index, kg/m 2.14 Mean SD Fat mass, kg.13 Mean SD Body fat, %.13 Mean SD Alcohol, g/d.27 Median Range Total energy intake, kj/d.31 Mean 7,818 8,096 SD 1,946 1,788 Physical activity, MET h/wk.36 Median Range Education.03 Primary school No. of patients 5 5 % 5 5 Technical/professional school No. of patients % Secondary school No. of patients % Academic degree No. of patients % Ever used hormone therapy for menopausal complaints.24 No. of patients % Abbreviations: SD, standard deviation; MET, metabolic equivalent. Differences between groups were analyzed with the following tests: t test for normal distributed continuous variables, Mann-Whitney U test for non-normally distributed continuous variables, and 2 test for binomial variables. Nutritional information of one woman is missing in the intervention group. MET hours per week spent on at least moderate-intensity (4 MET) activities. This was calculated using information from the Modified Baecke Questionnaire by American Society of Clinical Oncology JOURNAL OF CLINICAL ONCOLOGY

4 Exercise and Postmenopausal Sex Hormone Levels imputed at the lowest detectable level. Although SHBG is technically not a hormone, for reasons of convenience, it will be referred to as such. Physical Activity and Energy Intake Habitual exercise was measured by the validated Modified Baecke Questionnaire at baseline and 12 months. 25,26 We calculated the metabolic equivalent (MET) hours per week spent on at least moderate-intensity activities by coding the sports and leisure time activities reported in this questionnaire according to the Ainsworth Compendium of Physical Activities. 27,28 Moderate-intensity activities included all sports and leisure time activities of at least 4 MET based on the Dutch standard, which implies that moderate activities lie between 4 and 6.5 MET. 29 At baseline and 12 months, energy intake was measured by a food frequency questionnaire assessing dietary habits in the past month. 30,31 All completed questionnaires were checked by a dietician. Statistical Analysis We calculated that 90 participants per study arm were required to detect an intervention effect of 5.5 pg/ml estrone level with 90% power, including a dropout rate of 30%. Because all hormones showed skewed distributions, log-transformed values were used for data analyses. The primary trial analysis was performed according to the intent-to-treat principle. Between-group differences of sex hormone levels were analyzed with generalized estimating equations (GEEs). This longitudinal data analysis technique is suitable to investigate the course over time of the outcome variable and to compare this overall effect between study arms. In all models, the outcome variable (ie, one of the hormones measured at 0, 4, and 12 months) was analyzed as a dependent variable using study group as key independent variable adjusted for the baseline hormone measurement. Adjustment for baseline leads to equal starting points for both groups, and therefore, the intervention effect is presented Table 2. Hormone Concentrations, Total Body Fat, and Physical Activity Level at Baseline and 4 and 12 Months and Difference Between Intervention and Control Group Characteristic Baseline 4 Months 12 Months % Change From 0 to 4 Months % Change From 0 to 12 Months overall * 95% CI P overall * No. of patients Intervention Control Estrogens Free estradiol, pg/ml Intervention to Control Estradiol, pg/ml Intervention to Control Estrone, pg/ml Intervention to Control Estrone sulfate, ng/ml Intervention to Control Androgens Free testosterone, pg/ml Intervention to Control Testosterone, pg/ml Intervention to Control Androstenedione, pg/ml Intervention 1,146 1,097 1, to Control 1,172 1,181 1, SHBG, nmol/l Intervention to Control Body fat, % Intervention to 0.13 Control Physical activity level, median MET h/wk Intervention 4.9 NA to 18.3 Control 4.3 NA NOTE. Hormone concentrations were non-normally distributed, and therefore, all hormone values were log transformed. Presented are the geometric means of the hormones. Abbreviations: SHBG, sex hormone binding globulin; MET, metabolic equivalent; NA, not applicable. *The coefficient (and P value) represents the overall intervention effect on hormone change over time (adjusted for baseline) and was derived from a generalized estimating equation (GEE) model (coefficient on study group). Because the GEE models were based on log-transformed hormone data, the presented is the antilogarithm of the original. Therefore, the antilogarithm of the is a ratio that indicates whether the hormone level is, on average, higher in the intervention group compared with controls ( 1) or lower ( 1; eg, 0.9 indicates that the hormone level in the intervention group is on average 10% lower in the intervention group compared with the control group). Mean between-group difference (95% CI) in percent body fat (intervention control). MET hours per week spent on at least moderate-intensity (4 MET) activities. CI of this median difference was calculated using bootstrapping techniques by American Society of Clinical Oncology 4495

5 Monninkhof et al by the coefficient on study group. The group-time interaction provides information about whether the observed effect is stronger at the beginning or at the end of the study. 32 Because the GEE models were based on log-transformed hormone data, we presented the antilogarithm of the original coefficient in the tables. The antilogarithm of the is a ratio that indicates whether the hormone level is, on average, higher in the intervention group compared with controls ( 1) or lower ( 1; eg, 0.9 indicates that the hormone level in the intervention group is, on average, 10% lower in the intervention group compared with controls). To investigate the association between change in body fat (independent variable) and change in sex hormone levels (dependent variable), we also performed GEE analyses. Furthermore, we analyzed the intervention effect within the subgroup of women who lost more than 2% body fat. We chose this cutoff point to compare results with the study of McTiernan et al. 33,34 This study observed significant declines in sex hormone levels for women who lost more than 2% body fat. Additionally, we conducted GEE analyses with change in body fat as a covariate in the models. If body fat is an intermediate factor in the relation between physical activity and sex hormone levels, the intervention effect should attenuate when adjusting for change in body fat. We also performed a per-protocol analysis by excluding participants who were not compliant with the study protocol. Exercisers were considered not compliant if they missed 30% or more of all group sessions or participated in a formal weight loss program. Noncompliance in control women was considered as having started an exercise program or a formal weight loss program. RESULTS A total of 189 women were randomly allocated to either the intervention (n 96) or control group (n 93; Fig 1). At baseline, women in both groups were comparable with respect to age, alcohol use, fitness level, and androgen concentrations (Tables 1 and 2). However, despite block random assignment, baseline differences for several characteristics were observed, including education level, body fat, BMI, and estradiol levels (all higher for the control group, but only significant for education level). The number of dropouts in the SHAPE study was low (n 6; 3.2%) and included one woman in the intervention group and five women in the control group. The number of participants (intervention and control groups) who did not comply with the study protocol was 46 (24.3%). Thirty-five participants (36.5%) in the intervention group missed 30% or more of all group sessions, and 11 participants (11.8%) in the control group started exercise or a weight loss program. Adverse events as a result of the exercise program were not reported. Hormone measurements were available for all women at baseline, for 181 women (95.8%) at 4 months, and 182 women (96.3%) at 12 months. Self-reported level of moderate physical activity was higher at the end of the study, both for controls and exercisers. However, the activity level increased more in the intervention group (median: from 4.9 to 19.8 MET h/wk) than in controls (median: from 4.3 to 5.8 MET h/wk), and the difference between groups is significant. Furthermore, the intervention resulted in a significant higher loss of percent body fat compared with controls ( 0.43, ie, the mean percent body fat in the intervention group minus the mean percent body fat in the control group adjusted for baseline in the GEE analysis; Table 2). Also, the proportion of women who lost more than 2% fat was somewhat higher in the intervention group versus the control group (42% v 34%, respectively; P.27). Estrogens The intervention group compared with controls showed, on average, larger declines in levels of estradiol, estrone, and estrone sulfate ( range, 0.97 to 0.99), but these effects were not significant. The results on free estradiol were null (Table 2). In addition, among women who lost more than 2% body fat, declines in all estrogens were not significantly different between exercisers and controls (Table 3). Adjustment for change in body fat (total study population) attenuated the coefficients in the GEE analyses (Table 4). Furthermore, changes in body fat were significantly (positively) associated with changes in estrogen levels (Table 5); that is, fat loss resulted in a decline of estrogen levels. Per-protocol analysis showed similar results for the changes in estrogen levels in both groups (data not shown). Androgens On average, testosterone (free and total) and androstenedione showed larger declines in the intervention group compared with controls ( range, 0.97 to 0.99), although these results were not significant (Table 2). In women who lost more than 2% body fat, the androgen level was significantly more decreased in exercisers than in controls (Table 3). For free and total testosterone, the interaction with time was significant, indicating that the difference between the intervention and control groups among women who lost body fat was only present at 4 months. Adjustment for change in body fat (total study population) attenuated the intervention effect on testosterone (free and total; Table 4). Change in body fat was positively, but not significantly, associated with changes in androgen levels (Table 5). Per-protocol analysis showed similar results for the changes in free testosterone and androstenedione (data not shown). For testosterone, however, this analysis showed a significant decline in exercisers compared with controls, which occurred mainly in the first 4 months (exercisers: 8.4%; controls: 1.7%). SHBG During the study, SHBG levels were, on average, 2% lower in the intervention group compared with controls ( 0.98), although this was not significant (Table 2). The per-protocol analysis showed similar results (data not shown). In women who lost more than 2% body fat, SHBG levels were, on average, 5% lower in the intervention group compared with controls ( 0.95; P.09; Table 3). Adjustment for change in body fat (total study population) did not change the association (Table 4). Additionally, a change in body fat was not significantly associated with a change in SHBG levels (Table 5). DISCUSSION Overall, the exercise intervention of the SHAPE study did not result in favorable effects on sex hormone levels in sedentary, postmenopausal women. Among women who lost more than 2% body fat, declines in all estrogens were not significantly different between exercisers and controls. Androgen levels decreased significantly in the exercise group who lost body fat compared with their peers in the control group. Furthermore, this study confirmed that fat loss was significantly associated with declines in postmenopausal estrogen levels. Although not significant, a similar trend was observed for the androgens by American Society of Clinical Oncology JOURNAL OF CLINICAL ONCOLOGY

6 Exercise and Postmenopausal Sex Hormone Levels Table 3. Hormone Concentrations at Baseline and 4 and 12 Months and Difference Between Intervention and Control Groups for the Subgroup of Women Who Lost 2% Body Fat Hormone Baseline 4 Months 12 Months % Change From 0 to 4 Months % Change From 0 to 12 Months overall * 95% CI P overall * No. of patients Intervention Control Estrogens Free estradiol, pg/ml Intervention to Control Estradiol, pg/ml Intervention to Control Estrone, pg/ml Intervention to Control Estrone sulfate, ng/ml Intervention to Control Androgens Free testosterone, pg/ml Intervention to Control Testosterone, pg/ml Intervention to Control Androstenedione, pg/ml Intervention 1,118 1,003 1, to Control 1,299 1,308 1, SHBG, nmol/l Intervention to Control NOTE. Hormone concentrations were non-normally distributed, and therefore, all hormone values were log transformed. Presented are the geometric means of the hormones. Abbreviation: SHBG, sex hormone binding globulin. *The coefficient (and P value) represents the overall intervention effect on hormone change over time (adjusted for baseline) and was derived from a generalized estimating equation (GEE) model (coefficient on study group). Because the GEE models were based on log-transformed hormone data, the presented is the antilogarithm of the original. Therefore, the antilogarithm of the is a ratio that indicates whether the hormone level is, on average, higher in the intervention group compared with controls ( 1) or lower ( 1; eg, 0.9 indicates that the hormone level in the intervention group is on average 10% lower in the intervention group compared with the control group). Interaction with time in the generalized estimating equation analysis was statistically significant. The group-time interaction provides information about whether the observed intervention effect is stronger at the beginning or at the end of the intervention period. This study is the second intervention study assessing the effects of exercise on sex hormone levels in sedentary postmenopausal women. The first study of McTiernan et al 33,34 reported the effects of a 12-month moderate-intensity exercise intervention on sex hormones among 173 sedentary, overweight, postmenopausal women. In contrast to our study, that study did observe significant declines in serum estrogens in exercisers compared with controls, but only in women who lost more than 2% body fat. The decline in estrogen levels in the intervention group of that study compared with ours was even slightly smaller, but they observed no declines in the control group, whereas we did. This might partly be explained by the fact that control women, despite our request to maintain their usual lifestyle pattern, decreased energy intake and became (slightly) more physically active (Hertogh et al, submitted for publication). However, the contrast in exercise level between the intervention and control groups at the end of our study was still substantial. Another difference was the inclusion of overweight and obese women (mean BMI, 30.5 kg/m 2 ) in the study by McTiernan et al, 33,34 whereas we included women with a BMI between 22 and 40 kg/m 2 (mean BMI, 27.0 kg/m 2 ). Overweight women have more fat mass and, therefore, higher levels of estrogens and more potential for improvement. Body fat loss in the study by McTiernan et al was also larger than in the SHAPE study ( 1.4 v 0.81 kg, respectively). Furthermore, lack of an intervention effect might be explained by the amount of exercise. Women in the study by McTiernan et al were asked to exercise 3.75 hours a week compared with 2.5 hours in the SHAPE study. However, the intensity level of the SHAPE exercise program was higher, which might compensate for the shorter duration. Interestingly, effects on androgen levels were similar; there was no effect in the total intervention group but a significant decline for testosterone in women who lost more than 2% of body fat. 33 Also for androstenedione, both studies showed the same trend in favor of the exercise group among women who lost body fat, although the difference between exercise and control groups was not significant in the study by McTiernan et al. 33, by American Society of Clinical Oncology 4497

7 Monninkhof et al Table 4. Change of the Coefficients in the GEE Model When Adjusting for Change in Body Fat Hormone* Unadjusted Adjusted Change (%) Free estradiol Estradiol Estrone Estrone sulfate Free testosterone Testosterone Androstenedione SHBG NOTE. The outcome in each of the eight GEE models is sex hormone level measured at three occasions (baseline, 4 months, and 12 months), and covariates are study group and baseline hormone level. Abbreviations: GEE, generalized estimating equation; SHBG, sex hormone binding globulin. *Hormone concentrations were non-normally distributed and, therefore, log transformed. Thus, the s of these models correspond to the natural logarithm of the sex hormone in the model. The unadjusted is the coefficient on study group (intervention v control) from the GEE analysis. This coefficient indicates the log-transformed mean difference in hormone level between the intervention and control group. The adjusted coefficient is from the same model, adding percent change in body fat to the model. Change (adjusted beta unadjusted beta/unadjusted beta) 100%. Table 5. Associations Between Change in Percent Body Fat (continuous, independent variable) and Change in Sex Hormone Levels (continuous, dependent variable) Hormone 95% CI Estrogens Free estradiol to 0.02 Estradiol to 0.90 Estrone to 1.92 Estrone sulfate to 0.03 Androgens Free testosterone to 0.30 Testosterone to 0.01 Androstenedione to 0.02 SHBG to 0.78 Abbreviation: SHBG, sex hormone binding globulin. Generalized estimating equation analysis with two measurements of change ( baseline and 4 months and 4 and 12 months) of both independent and dependent variable. The is the coefficient of change in hormone level per unit (%) change in percent body fat (eg, as mean percent body fat increases by 1 unit the mean estrone level increases by 1.01 pg/ml). The results of the subgroup analysis among women who lost body fat should be considered with caution because it is a post hoc analysis. A further limitation is that we deviated from protocol by not presenting the results on abdominal fat measured by ultrasound because this measure seemed to be not valid. We expected the measurement error to be larger than the anticipated differences. Furthermore, the aim of the SHAPE study was to assess the effect of exercise on hormone levels and not to establish weight loss. However, we cannot rule out that women participated in our study with the aim to lose weight. If so, this might explain the larger decrease in energy intake among women allocated to the control group ( 445 kj/d) compared with women in the intervention group ( 27 kj/d). Adjustment for energy intake did not change the results. Mean daily alcohol intake also decreased but in the same amount in intervention and control women ( 1.3 and 1.4 g, respectively). Use of medication was low and not different between groups (data not shown). Strengths of this study include the relatively large study population with a compliance rate comparable to other exercise intervention trials in older adults. 35 The combined aerobic and strength training comprised an exercise level achievable by postmenopausal women. The contrast in the level of physical activity after 12 months was substantial between the intervention and control groups. Furthermore, comprehensive measurement of body composition with dualenergy x-ray absorptiometry allowed us to address the effect of hormone changes in relation to body fat. In conclusion, the exercise intervention in the SHAPE study did not result in significant decreases in serum estrogens and androgens in sedentary, postmenopausal women. Androgen levels were significantly more decreased in exercisers than in controls only in a subgroup of women who lost more than 2% body fat. Furthermore, this study confirmed that fat loss is associated with changes in postmenopausal sex hormone levels. AUTHORS DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST The author(s) indicated no potential conflicts of interest. AUTHOR CONTRIBUTIONS Conception and design: Evelyn M. Monninkhof, Petra H.M. Peeters, Albertine J. Schuit Financial support: Evelyn M. Monninkhof, Petra H.M. Peeters, Albertine J. Schuit Collection and assembly of data: Evelyn M. Monninkhof, Petra H.M. Peeters, Albertine J. Schuit Data analysis and interpretation: Evelyn M. Monninkhof, Miranda J. Velthuis, Petra H.M. Peeters, Jos W.R. Twisk, Albertine J. Schuit Manuscript writing: Evelyn M. Monninkhof Final approval of manuscript: Miranda J. Velthuis, Petra H.M. Peeters, Jos W.R. Twisk, Albertine J. Schuit REFERENCES 1. Monninkhof EM, Elias SG, Vlems FA, et al: Physical activity and breast cancer: A systematic review. Epidemiology 18: , Lahmann PH, Friedenreich C, Schuit AJ, et al: Physical activity and breast cancer risk: The European Prospective Investigation into Cancer and Nutrition. Cancer Epidemiol Biomarkers Prev 16:36-42, Bernstein L, Ross RK, Lobo RA, et al: The effects of moderate physical activity on menstrual cycle patterns in adolescence: Implications for breast cancer prevention. Br J Cancer 55: , WHO: Weight Control and Physical Activity. Lyon, France, IARC Press, Key T, Appleby P, Barnes I, et al: Endogenous sex hormones and breast cancer in postmenopausal women: Reanalysis of nine prospective studies. J Natl Cancer Inst 94: , Tamimi RM, Byrne C, Colditz GA, et al: Endogenous hormone levels, mammographic density, and subsequent risk of breast cancer in postmenopausal women. J Natl Cancer Inst 99: , Cauley JA, Gutai JP, Kuller LH, et al: The epidemiology of serum sex hormones in postmenopausal women. Am J Epidemiol 129: , Kaye SA, Folsom AR, Soler JT, et al: Associations of body mass and fat distribution with sex hormone concentrations in postmenopausal women. Int J Epidemiol 20: , Nelson ME, Meredith CN, Dawson-Hughes B, et al: Hormone and bone mineral status in by American Society of Clinical Oncology JOURNAL OF CLINICAL ONCOLOGY

8 Exercise and Postmenopausal Sex Hormone Levels endurance-trained and sedentary postmenopausal women. J Clin Endocrinol Metab 66: , McTiernan A, Wu L, Chen C, et al: Relation of BMI and physical activity to sex hormones in postmenopausal women. Obesity (Silver Spring) 14: , Verkasalo PK, Thomas HV, Appleby PN, et al: Circulating levels of sex hormones and their relation to risk factors for breast cancer: A cross-sectional study in 1092 pre- and postmenopausal women (United Kingdom). Cancer Causes Control 12:47-59, Huang Z, Hankinson SE, Colditz GA, et al: Dual effects of weight and weight gain on breast cancer risk. JAMA 278: , Vainio H, Kaaks R, Bianchini F: Weight control and physical activity in cancer prevention: International evaluation of the evidence. Eur J Cancer Prev 11:S94-S100, 2002 (suppl 2) 14. Key TJ, Allen NE, Verkasalo PK, et al: Energy balance and cancer: The role of sex hormones. Proc Nutr Soc 60:81-89, Judd HL, Shamonki IM, Frumar AM, et al: Origin of serum estradiol in postmenopausal women. Obstet Gynecol 59: , Hankinson SE, Willett WC, Manson JE, et al: Alcohol, height, and adiposity in relation to estrogen and prolactin levels in postmenopausal women. J Natl Cancer Inst 87: , Lukanova A, Lundin E, Zeleniuch-Jacquotte A, et al: Body mass index, circulating levels of sexsteroid hormones, IGF-I and IGF-binding protein-3: A cross-sectional study in healthy women. Eur J Endocrinol 150: , McTiernan A, Ulrich CM, Yancey D, et al: The Physical Activity for Total Health (PATH) Study: Rationale and design. Med Sci Sports Exerc 31: , Pate RR, Pratt M, Blair SN, et al: Physical activity and public health: A recommendation from the Centers for Disease Control and Prevention and the American College of Sports Medicine. JAMA 273: , Monninkhof EM, Peeters PH, Schuit AJ: Design of the Sex Hormones and Physical Exercise (SHAPE) study. BMC Public Health 7:232, Burger H, Woods NF, Dennerstein L, et al: Nomenclature and endocrinology of menopause and perimenopause. Expert Rev Neurother 7:S35-S43, 2007 (suppl 11) 22. Boker LK, van Noord PA, van der Schouw YT, et al: Prospect-EPIC Utrecht: Study design and characteristics of the cohort population European Prospective Investigation into Cancer and Nutrition. Eur J Epidemiol 17: , Rinaldi S, Dechaud H, Biessy C, et al: Reliability and validity of commercially available, direct radioimmunoassays for measurement of blood androgens and estrogens in postmenopausal women. Cancer Epidemiol Biomarkers Prev 10: , Rinaldi S, Geay A, Dechaud H, et al: Validity of free testosterone and free estradiol determinations in serum samples from postmenopausal women by theoretical calculations. Cancer Epidemiol Biomarkers Prev 11: , Voorrips LE, Ravelli AC, Dongelmans PC, et al: A physical activity questionnaire for the elderly. Med Sci Sports Exerc 23: , Hertogh EM, Monninkhof EM, Schouten EG, et al: Validity of the Modified Baecke Questionnaire: Comparison with energy expenditure according to the doubly labeled water method. Int J Behav Nutr Phys Act 5:30, Ainsworth BE, Haskell WL, Whitt MC, et al: Compendium of physical activities: An update of activity codes and MET intensities. Med Sci Sports Exerc 32:S498-S504, 2000 (suppl 9) 28. Ainsworth BE, Haskell WL, Leon AS, et al: Compendium of physical activities: Classification of energy costs of human physical activities. Med Sci Sports Exerc 25:71-80, Kemper HC, Ooijendijk WTM, Stiggelbout M: Consensus about the Dutch recommendation for physical activity to promote health [Consensus over de Nederlandse norm gezond bewegen]. Tijdschr Soc Geneeskd 78: , Feunekes IJ, van Staveren WA, Graveland F, et al: Reproducibility of a semiquantitative food frequency questionnaire to assess the intake of fats and cholesterol in The Netherlands. Int J Food Sci Nutr 46: , Feunekes GI, van Staveren WA, De Vries JH, et al: Relative and biomarker-based validity of a food-frequency questionnaire estimating intake of fats and cholesterol. Am J Clin Nutr 58: , Twisk JWR: Analysis of experimental studies, in Twisk JWR (ed): Applied Longitudinal Data Analysis for Epidemiology. Cambridge, United Kingdom, Cambridge University Press, 2003, pp McTiernan A, Tworoger SS, Rajan KB, et al: Effect of exercise on serum androgens in postmenopausal women: A 12-month randomized clinical trial. Cancer Epidemiol Biomarkers Prev 13: , McTiernan A, Tworoger SS, Ulrich CM, et al: Effect of exercise on serum estrogens in postmenopausal women: A 12-month randomized clinical trial. Cancer Res 64: , Martin K, Sinden A: Who will stay and who will go? A review of older adults adherence to randomized controlled trials of exercise. J Aging Phys Act 9:91-114, by American Society of Clinical Oncology 4499